What Is a Browser Tab, and What It Cannot Separate

The question usually arrives sideways. Someone is looking at a browser window holding forty tabs, or trying to work out why two tabs signed into different accounts keep interfering with each other, or reading a support article that says to open a link in a new tab as though the meaning were obvious. The word is used constantly and defined almost never.

A short definition works for most of it. A browser tab is one loaded document plus the state that belongs to it, presented as a switchable strip in a single window. The useful part is what follows from that definition, because the things a tab does not do are the reason a Mac full of them starts to feel unmanageable.

A tab is a view, not a container

The strip at the top of the window is the interface. Underneath it sits a document: an address, the page loaded from it, the scroll position, the form fields partly filled, the history of what was visited in that strip, and the JavaScript still running.

That bundle belongs to the tab. Close the tab and it goes, which is why reopening from history restores the page but not the half-written comment.

What does not belong to the tab is everything that identifies the person using it. Cookies, local storage, saved passwords, and installed extensions belong to the browser profile, one level up. Every tab in a profile draws from the same pool. This single fact explains the most common surprise in multi-account work: opening a second tab and signing into a different account does not give two independent sessions, it replaces or stacks the session in the shared pool, and both tabs see the result.

The strip itself carries very little information, and it is worth knowing exactly how little. A tab shows a site icon, a title the page chooses for itself, and an indicator when audio is playing. The title is not verified, so two tabs on different sites can present the same words. As the count rises, the title is truncated first and then dropped entirely, leaving a row of icons that are frequently the only thing left to distinguish one service from another. Everything else that would help, such as which account is signed in or whether anything has changed since the last look, is inside the page rather than on the strip.

Private windows are the exception that proves the shape. They work not because they are tabs but because they are a separate storage container, discarded when the last window in it closes. The isolation comes from the container, never from the tab.

What one open tab costs

Tabs are not free, and the reason is architectural rather than accidental. Chromium's design documentation describes the model directly:

In general, each new window or tab opens in a new process. Source: chromium.org

Separate processes are what stop one broken page from taking the browser down with it, and what let the browser enforce a boundary between sites. They also mean that each tab carries its own share of memory rather than sharing one pool.

The same document notes that the browser does not scale this indefinitely. Chromium has strategies to assign new tabs to existing processes when the total number of processes grows too large, and it can release the working set of a background tab so the system swaps that memory out first. Modern browsers extend this with discarding, where a background tab is unloaded and reloaded on return.

The practical consequence is a shape, not a number. Memory pressure does not rise smoothly with tab count. It stays flat while background tabs are cheap, then rises when tabs are active, holding a live connection, or running a heavy web app. Forty documentation pages sitting idle cost far less than six chat and video tools that never stop talking to a server. Counting tabs is therefore a poor diagnostic. Counting the ones that never get closed is a better one.

What tabs cannot express

The limits show up when the same browser holds work that belongs to different contexts.

Identity. Two tabs cannot hold two accounts for the same service independently, because the session lives in the profile. The workarounds all move up a level: a second profile, a second browser, or a container feature that gives a tab its own storage.

Attention. A tab has no state for how urgent it is. A build log and a payroll form look identical in the strip, and both shrink to a favicon at the same rate as the count rises.

Reachability from outside. This is the one that costs the most time on a Mac. The operating system schedules applications and windows. It does not know what a tab is. Command+Tab reaches the browser, not the inbox inside it. The Dock shows one icon regardless of how many services are open. Notification grouping attributes every alert to the browser, so a message from a client and a marketing email arrive as the same kind of event from the same sender.

That last point is why tab counts keep climbing. A tab that cannot be reached directly is a tab that is easier to leave open than to find again later. The pile is a rational response to a missing addressing mechanism.

What survives closing a tab, and what does not

Closing is where the boundary of a tab becomes visible, because some things come back and some do not.

What comes back is anything the browser records at the profile level. The address is in history. The login stays valid because the cookie was never in the tab. Reopening the most recently closed tab restores the page and its back-forward history, and a browser set to continue where it left off will bring the whole set back at the next launch.

What does not come back is the state the page was holding in memory at that moment. Text typed into a field that was never submitted, the position inside a long document, an unsaved draft in an editor that does not autosave, a video paused halfway, a filtered view of a dashboard built up over several clicks. Some sites save these to local storage and restore them, and that is a decision each site makes rather than a browser feature.

The gap between those two lists is why tabs accumulate. Closing a tab is cheap for pages that hold no state and expensive for tools that hold a lot of it, and the strip gives no signal about which is which. Given that uncertainty, leaving everything open is the rational choice, and the count grows.

Telling an expensive tab from a cheap one

Chrome ships a task manager that lists every tab, extension, and internal process with its memory footprint and processor share. On a Mac it opens from the Window menu of the browser rather than from a keyboard shortcut. Sorting by memory usually reveals that a small number of entries account for most of the total, and that the entries are rarely the ones a person would guess.

Three patterns show up consistently. Tools holding a live connection to a server, such as chat and collaborative editors, stay expensive whether or not anyone is looking at them. Pages with autoplaying or embedded video hold decoding work and can keep a processor busy in the background. Extensions appear in the same list, and one poorly behaved extension can outweigh several dozen ordinary tabs.

The practical use of that list is not to close things. It is to identify which tabs are actually applications. Anything that stays near the top of the memory list day after day is a tool being used, not a page being read, and it is the category that benefits from being moved out of the strip entirely.

What browsers have added on top of tabs

Rather than replace the unit, browser vendors have layered organisation onto it. Chrome now offers several mechanisms at once, and they solve different problems.

Mechanism What it groups Persists after closing Known catch
Pinned tabs Single pages kept at the left Yes, per window Only a size and position change
Tab groups Named, coloured sets of tabs Yes, and syncs to other devices Deleting removes it everywhere
Vertical tabs The whole strip, moved to the side Layout preference only Layout change, not isolation
Tab search Nothing, it finds instead Not applicable Finding is not organising
Separate profiles Everything, including storage Yes All windows are one macOS app

Two of those rows deserve a closer look. Vertical tabs became a standard Chrome feature in 2026, with Google describing the activation as a right click on any Chrome window followed by Show Tabs Vertically, and the benefit as being able to read full page titles even when the tab count hits double digits. It changes legibility, which is real, and changes nothing about isolation.

Tab groups carry a sharper edge. Google's documentation states that when a tab group is deleted it is removed on that device and on other devices using the same Google Account, and that ungrouping tabs also deletes the group across devices. A synced group is convenient until a tidy-up on a laptop removes a set that a desktop was still relying on.

When the tab stops being the right unit

The definition at the top holds well for browsing. A tab is an excellent unit for a document that is read and then closed. Search results, documentation, an article, a form: all of these have a beginning and an end, and the tab ends with them.

It holds badly for software. A chat tool, a mail client, a project board, and a terminal in a browser are not documents. They stay open for months, hold identity, produce notifications, and are switched between dozens of times a day. Treating them as documents means every one of them competes for the same strip, inherits the same session, and hides behind the same Dock icon.

The distinction to draw is not between many tabs and few. It is between pages being read and applications being used. Once those two are separated, the browsing pile can stay a pile without doing any harm, and the applications can move to a unit that gives each one its own window, its own storage, and its own place in Command+Tab. That is what a browser built around workspaces provides, and the Workspaces page sets out how the unit is defined. The Supported apps list shows which services are already configured to run that way.

The change worth making first

Sort the open tabs once into two lists: pages that will be closed today, and tools that have been open for more than a week. Leave the first list alone. Move the second list out of the strip, giving each tool its own window and its own storage, which is the arrangement SpaceDeck is built around. The tab count matters far less afterwards, because nothing important is hiding in it.

Frequently asked questions

What is the difference between a browser tab and a browser window?

A window is what the operating system manages, with its own place in Command+Tab and its own entry in the Dock. A tab is a document inside that window, invisible to macOS. That difference is why a tab cannot be reached with a keyboard shortcut at the system level while a window can.

Do open tabs use memory even when they are not visible?

Yes, though less than active ones. Chromium generally opens each tab in its own process and can release a background tab's working set so the system swaps that memory out first. Browsers may also discard a background tab entirely and reload it on return, which trades memory for a short delay.

Does opening a second tab let two accounts stay signed in?

Not by itself. Cookies and storage belong to the browser profile rather than the tab, so all tabs in a profile share one session pool. Holding two accounts open independently requires a separate profile, a separate browser, or a feature that gives each space its own storage.

Is there a maximum number of tabs a browser can open?

No fixed limit is published, and the practical ceiling depends on what the tabs contain rather than how many there are. Idle text pages cost little, while chat, video, and dashboard tools hold live connections and cost far more. Available memory and the browser's discarding behaviour set the real boundary.

Back to all posts